심방세동 진단을 위한 시계열성을 보전한 2차원 심전도 데이터 기반의 합성곱 신경망 모델

국문 초록 (Abstract)

본 논문은 심장의 이상징후 중 심방세동 진단을 위하여 시계열성을 보전한 2D 심전도 데이터 기반의 합성곱 신경망 모델을 제안한다. 제안 모델은 시계열성을 보전한 2D 구조의 심전도 데이터를 사용하여 1D 합성곱 신경망 모델이나 기존의 2D 합성곱 신경망 모델에 비하여 심방세동 진단 성능을 높인다. 제안 방법의 유효성을 검증하기 위하여 Physionet 2017에 공개된 심전도 데이터를 활용하여 비교실험을 수행하였다. 1D 합성곱 신경망을 사용한 모델, 기존의 2D 합성곱 신경망 모델들과 제안 방법을 동일 환경에서 비교한 결과, 제안 방법이 다른 방법들에 비하여 더 적은 수의 학습 파라미터를 사용하면서도 더 높은 F1 점수와 정밀도(precision)를 보임을 확인하였다.

다국어 초록 (Multilingual Abstract)

This paper proposes a convolutional neural network model based on 2D electrocardiogram data that preserves time series for diagnosing of atrial fibrillation among cardiac abnormalities. The proposed model uses ECG data of a 2D structure that preserves time series to improve atrial fibrillation diagnosing performance compared to the 1D convolutional neural network model and the existing 2D convolutional neural network model. To confirm the superiority of the proposed method, a comparative experiment was conducted using the ECG data published in Physionet 2017. A comparing of a 1D convolutional neural network model, two existing 2D convolutional neural network models and ours in the same experimental environments, the proposed method shows a higher F1 score and precision with fewer learning parameters.

목차 (Table of Contents)

• 요약
• Abstract
• 1. 서론
• 2. 관련 연구
• 3. 시계열성을 보전한 2차원 심전도 데이터 기반의 합성곱 신경망 모델
• 4. 실험 및 결과고찰
• 5. 결론 및 향후 연구
• References

참고문헌 (Reference)

1 정보영, "심방세동 치료 가이드라인" 대한의사협회 62 (62): 265-274, 2019

2 R. M. John, "Ventricular Arrhythmias and Sudden Cardiac Death" 380 (380): 1520-1529, 2012

3 M. P. Turakhia, "Rationale and Design of a Large-scale, App-based Study to Identify Cardiac Arrhythmias using a Smartwatch: the Apple Heart Study" 66-75, 2019

4 A. Fujiki, "QT/RR Relation During Atrial Fibrillation based on a Single Beat Analysis in 24-h Holter ECG:The Role of the Second and Further Preceding RR Intervals in QT Modification" 57 (57): 269-274, 2011

5 S. Bach, "On Pixel-wise Explanations for Non-linear Classifier Decisions by Layer-wise Relevance Propagation" 10 : 2015

6 A. Krizhevsky, "Imagenet Classification with Deep Convolutional Neural Networks" 60 (60): 84-90, 2017

7 S. Datta, "Identifying Normal, AF and Other Abnormal ECG Rhythms using a Cascaded Binary Classifier" 2017

8 P. Li, "High-performance Personalized Heartbeat Classification Model for Long-term ECG Signal" 64 (64): 78-86, 2016

9 S. Hong, "ENCASE: an ENsemble ClASsifiEr for ECG Classification using Expert Features and Deep Neural Networks" 2017

10 F. Andreotti, "Comparing feature-based Classifiers and Convolutional Ceural Cetworks to Detect Arrhythmia from Short Segments of ECG" 2017

11 P. Rajpurkar, "Cardiologist-level Arrhythmia Detection with Convolutional Neural Networks" 25 : 65-69, 2019

12 U. R. Acharya, "Automated Detection of Arrhythmias using Different Intervals of Tachycardia ECG Segments with Convolutional Neural Network" 405 : 81-90, 2017

13 Atrial Fibrillation Guidelines Committee, "Atrial Fibrallation"

14 T. Teijeiro, "Arrhythmia Classification from the Abductive Interpretation of Short Single-lead ECG Records" 2017

15 G. D. Clifford, "AF Classification from a Short Single Lead ECG Recording: the PhysioNet/Computing in Cardiology Challenge 2017" 2017

16 J. Kim, "A Novel Recommendation Approach based on Chronological Cohesive Units in Content Consuming Logs" 470 : 141-155, 2019

17 Y. Wu, "A Comparison of 1-D and 2-D Deep Convolutional Neural Networks in ECG Classification"